WAN 2.5 and 2.6 on VdoBloom: Flash, Text and Image to Video

VdoBloom runs four WAN video models: WAN 2.6 Text-to-Video, WAN 2.6 Image-to-Video, WAN 2.6 Flash and WAN 2.5 Image-to-Video. The 2.6 tiers are the ones to reach for, because they are cheaper per second and cover 5, 10 and 15 second clips at 720p and 1080p, while Flash adds native audio and nine duration steps from 2 to 15 seconds. The one WAN 2.5 model available is image-to-video only — there is no WAN 2.5 text-to-video option to pick.

Updated October 2026.

The four WAN models on VdoBloom

Alibaba’s WAN line arrives on VdoBloom as four separate models rather than one model with a version switch, so the version you want is a decision you make in the picker before you type a prompt. Two of them start from an image, one starts from text, and one is built for speed.

WAN 2.6 Flash — image-to-video, 2 to 15 seconds, 720p or 1080p, with native audio. Nine duration steps: 2, 3, 4, 5, 6, 8, 10, 12 and 15 seconds.
WAN 2.6 Text-to-Video — text-only clips at 5, 10 or 15 seconds, 720p or 1080p.
WAN 2.6 Image-to-Video — animates the exact image you upload, 5, 10 or 15 seconds, 720p or 1080p.
WAN 2.5 Image-to-Video — a still into a 5 or 10 second clip at 720p or 1080p. It is the older and the more expensive of the image-to-video options here.

WAN 2.6 Text-to-Video and WAN 2.6 Flash carry VdoBloom’s permissive content badge, which is the platform’s more flexible filtering tier. If you are unsure which of the four fits your clip, the honest short answer is that Flash is the fastest and cheapest way to move a photo, WAN 2.6 Text-to-Video is the only one that works without an image, and WAN 2.5 Image-to-Video is there for continuity with older projects rather than because it does something the 2.6 models cannot.

What each model actually costs

Every WAN model on VdoBloom is priced in credits, and the cost scales with both the resolution and the length of the clip. The price is shown before you generate, so you can check it against your balance rather than discovering it afterwards.

Model	Starts from	Durations	720p cost	1080p cost
WAN 2.6 Flash	Image	2 to 15s (9 steps)	4 credits/second (8 at 2s, 60 at 15s)	6 credits/second (12 at 2s, 90 at 15s)
WAN 2.6 Text-to-Video	Text	5, 10, 15s	35 / 75 / 120	60 / 125 / 190
WAN 2.6 Image-to-Video	Image	5, 10, 15s	35 / 75 / 120	60 / 125 / 190
WAN 2.5 Image-to-Video	Image	5, 10s	65 / 130	115 / 230

The comparison that matters most is the image-to-video one. WAN 2.6 Image-to-Video works out at roughly 7 credits per second at 720p, and WAN 2.5 Image-to-Video at roughly 13 credits per second at 720p — close to double for the same five seconds of video. At 1080p the gap is similar: about 12.5 credits per second on 2.6 against about 23 on 2.5. Flash sits below both at about 4 credits per second at 720p and 6 at 1080p, which is why it is the model to reach for when you are still working out what the shot should be.

Credits come from a subscription or from one-time packs that never expire, so a clip you generate in October and never use costs you nothing further. New accounts start with 10 free credits and no card, which covers a two-second WAN 2.6 Flash clip at 720p. Paid output is watermark-free with commercial rights, and full plan detail is on the pricing page.

WAN 2.6 Flash: the one to start with

Flash is the only WAN model here with nine duration steps, and that granularity is the point. A 2-second clip is often enough to test whether a still has the right amount of motion, and being able to step 2, 3, 4, 5 and 6 seconds means you can add motion in small increments instead of committing to a 5 or 10 second block. Its native audio also means the clip arrives with sound rather than needing a separate pass.

The trade-off is that Flash is image-to-video only. You need a still to begin with, whether that is a photograph, a generated image, or a frame you exported from an earlier clip.

WAN 2.6 Text-to-Video and Image-to-Video

These two share a duration ladder and a price table, which makes moving between them cheap in the sense that matters: the same 5-second 720p clip costs 35 credits whichever way you start. The difference is entirely in what the model has to invent.

Text-to-Video builds the whole frame from your words, so it controls composition, subject and lighting. Image-to-Video animates the exact uploaded image, which means your composition and your subject’s likeness are preserved and the model is only deciding how they move. If the shot depends on a specific person, product or location looking a particular way, that is an image-to-video job. If it depends on a scene that does not exist yet, it is a text-to-video job.

Both reach 15 seconds, which is long for a single generation and worth using deliberately: a 15-second take at 720p costs 120 credits, and if you only needed 10 seconds you have spent 45 credits on footage you will cut. Start shorter than you think you need.

WAN 2.5 Image-to-Video: when to use it

The 2.5 model animates a still at 5 or 10 seconds, at 720p or 1080p, for more credits than the equivalent 2.6 run. There is no configuration on it that the 2.6 models do not offer, so the reason to choose it is continuity: matching the look of clips you already made, or reproducing a result from a project that was built on it. For new work, WAN 2.6 Image-to-Video or Flash will get you there for less.

Six prompts that suit the WAN models

WAN models respond best to a clear subject, one explicit camera instruction, and a stated lighting condition. Six starting points:

Product turntable (image-to-video, Flash, 4s): “The bottle rotates slowly in place on a dark reflective surface, a single soft key light sweeping across the label, camera locked off, no other movement.”
Portrait breathing shot (image-to-video, Flash, 3s): “The subject turns their head slightly toward the light and blinks once, a slow locked-off push in, shallow depth of field, natural window light from the left.”
City establishing shot (text-to-video, 720p, 5s): “A rain-slicked city street at night, neon reflections in the puddles, a slow crane up from street level, pedestrians blurred in the background, cool blue and magenta lighting.”
Food reveal (image-to-video, 720p, 5s): “Steam rises from the dish as the camera drifts slowly to the right, warm overhead kitchen light, the plate stays perfectly still, handheld but stable.”
Landscape timelapse feel (text-to-video, 1080p, 10s): “Clouds move quickly over a mountain ridge at golden hour, long shadows sliding across the valley, a very slow dolly forward.”
Fashion walk (image-to-video, 1080p, 5s): “The model walks three steps toward the camera, coat moving naturally, camera tracks back at the same pace, broad daylight, no camera shake.”

For the underlying principles behind these — how much to specify, and what to leave to the model — they apply across every image-to-video model on the platform; see AI image-to-video best practices. If your clips are coming out as near-static images with a slight wobble, the fixes are different and are covered in how to make AI video not look like a slideshow.

Which WAN model to pick

Testing whether a still moves well: WAN 2.6 Flash at 2 or 3 seconds, 720p. Cheapest way to find out.
A photo that needs to move, at final quality: WAN 2.6 Image-to-Video at 1080p, 5 seconds.
A scene that does not exist yet: WAN 2.6 Text-to-Video at 720p, 5 seconds.
A clip with sound and fine duration control: WAN 2.6 Flash.
Matching an existing 2.5 project: WAN 2.5 Image-to-Video, accepting the higher per-second rate.

Limits worth knowing before you spend

No WAN 2.5 text-to-video. If you want text-only WAN output, it is the 2.6 model or nothing.
No aspect-ratio selector. The WAN video models follow the input image, and the text-to-video model does not expose a ratio control the way some other families do.
720p and 1080p only. There is no 4K tier in this family — for 4K output you need a different model.
Flash needs an image. It cannot start from a prompt alone.
Cost rises with both length and resolution at once. A 15-second 1080p run is the most expensive thing here; work up to it.

Frequently asked questions

Is there a WAN 2.5 text-to-video model on VdoBloom?

No. The only WAN 2.5 model on the platform is image-to-video, at 5 or 10 seconds in 720p or 1080p. Text-to-video in this family starts at WAN 2.6.

What is the cheapest WAN model on VdoBloom?

WAN 2.6 Flash, at about 4 credits per second at 720p. A 2-second test clip costs 8 credits, which is inside the 10 free credits a new account starts with.

Why is WAN 2.5 more expensive than WAN 2.6?

It is the older tier and is priced above the newer models — roughly 13 credits per second at 720p against about 7 on WAN 2.6 Image-to-Video. There is no feature on it that the 2.6 models lack, so it is best treated as a continuity option.

Which WAN model has audio?

WAN 2.6 Flash generates native audio with the clip. The other three produce video only.

Can I try WAN before paying?

Yes. New accounts get 10 free credits with no card required, which is enough for a 2-second 720p clip on WAN 2.6 Flash or a 1K image on the image models.

WAN 2.6 Image-to-Video and WAN 2.5 Image-to-Video both have their own model pages in the model catalogue, and the rest of the family sits beside them. For a broader comparison of the WAN tier against the current ByteDance models, read the WAN 3.0 guide.